Yongge Sun *, Guoying Sheng, Ping'an Peng, Jiamo Fu

Transcription

1 Organic Geochemistry 31 (2000) 1349± Compound-speci c stable carbon isotope analysis as a tool for correlating coal-sourced oils and interbedded shale-sourced oils in coal measures: an example from Turpan basin, north-western China Yongge Sun *, Guoying Sheng, Ping'an Peng, Jiamo Fu SKLOG, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou , PR China Abstract Molecular geochemical methods have shown that it is often di cult to di erentiate between coal- and interbedded shale-sourced oils, even though coals and interbedded shales may exhibit considerable organic in ux variation (e.g. land plant vs algal organic matter) due to the changes of depositional setting. However, compound-speci c stable carbon isotopic compositions are sensitive to the source input variations. Typically, speci c molecules are more depleted in 13 C with increasing content of aqueous biota. This hypothesis is examined and exempli ed by comparing the stable carbon isotopic ratios of n-alkanes from source rock extracts and related oils of the Turpan basin, northwestern China. Stable carbon isotopic values of n-alkanes extracted from coals and interbedded shales show that d 13 C values of n-alkanes with less than 20 carbon atoms vary only slightly. However, there are dramatic changes in the isotopic compositions of higher molecular weight n-alkanes. Furthermore, n-alkanes from coal extracts are enriched in 13 C relative to that of interbedded shales with excursions up to 2±3%. This comparison enables the di erentiation of coal- and interbedded shale-sourced oils, and provides information useful in assessing the hydrocarbon system of a basin. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Compound speci c stable carbon isotope analysis; Coal-sourced oils; Interbedded shale-sourced oils; Productivities of biota; Turpan basin; China 1. Introduction Recent work by Isaksen et al. (1998), Inan et al. (1998) and Killops et al. (1998) have unraveled the key controls on the oil and gas potential of humic coals. Isaksen et al. (1998) showed that the concentration of long-chained aliphatic hydrocarbons in the coal matrix is the main control on the oil potential of humic coals, and that a ``bulk'' measure of liptinite content may prove misleading when assessing the oil potential. However, questions remain as to how to characterize oils derived from coals. Powell and Boreham (1994) argued that the composition of coal-sourced oils represented the average of the hydrocarbons expelled from * Corresponding author. Fax: address: ygsun@gig.ac.cn (Y. Sun). the drainage area of a structure, which may not be represented by a single sample or a group of samples due to the strong heterogeneity in organic matter content and composition in coal-bearing sequence. Accordingly, It is di cult to pin the source of this type of oil down to a speci c stratigraphic intervals or particular lithofacies. A key problem is that common molecular geochemistry correlation techniques may not di erentiate between coals and associated shales containing dispersed organic matter from terrigenous higher plants. The traditional ``correlation'' of this type of oil consists of (i) the similarity between the heavy n-para n distribution of source extracts and produced crude oils (Gordon, 1985), (ii) speci c biomarker compound classes (e.g. sesqui-, di-, tetra-, and pentacyclic terpanes), which can be linked back to land plant communities (e.g. Philp, 1994; Isaksen, 1995). Although similarities may exist between crude oils and potential source rock /00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S (00)

2 1350 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349±1362 extracts in the biomarker assemblages, it does not support a genetic relationship because both coals and interbedded shales are usually the source beds (Hunt,1991), and they often have the same or similar biomarker distributions. One exception to this was presented by Curiale and Lin (1991). In their study, three distinctly di erent organic facies were identi ed among coals and shales in a deltaic setting. By comparing the pyrolysis-gas chromatography results of a collection of coals, Katz et al. (1991) concluded that the shales associated with the coals are possibly the primary source rocks for the most petroleum systems in coal-bearing basin in the southeastern Asian and southwestern Paci- c region, particularly in the Gippsland basin (Brooks,1970; Curry et al., 1994) and in central Sumatra (Williams et al., 1985). Since the petroleum reserves of a basin strongly rely upon the source rock attributes, including organic richness, type of organic matter, source rock volume (areal extent and stratigraphic thickness), level of thermal maturity and expulsion e ciency, it is necessary and useful to develop new methods to distinguish between coal-derived oils and interbedded shale-sourced oils. Such methods will allow more reliable assessment of the source rock potential of coal-bearing basins as well as reserve estimations. To date, much of the literature data concerning the origins of these types of oils is based on biomarker correlation techniques. However, the use of stable carbon isotopic analysis of bulk and individual compounds has shown to complement the molecular approach, and in certain instances to lead greater insights in oil-oil and oil-source correlation in terrigenous sequences (Boreham and Summons, 1999). Earlier work on the isotopic characterization of crude oils focused on the analysis of the total carbon or compound-type fractions (saturate, aromatic, polar and asphaltene) of crude oils, particularly on the use of stable carbon isotope type-curves (Stahl, 1978; Schoell, 1983; Weston et al., 1988; Clayton,1991). Given that both coals and interbedded shales can be the source for oil in coal-bearing sequence, the stable carbon isotopes of bulk kerogen or compound classes do not provide the required resolution due to the slight variation of organic matter types between coals and interbedded shales. As a support of this contention, Zhang (1997) compared the stable carbon isotopes of kerogen, total solvent extracts and bulk compound classes for coals and interbedded shales from Jurassic coal measures in the Turpan basin, north-western China. The study showed that the carbon isotopic compositions were similar between coal and interbedded shale (e.g. an average of 23.5% vs 23.8% for kerogens, and 26.6% vs 26.0%, for extracts respectively). On the other hand, coals and interbedded shales in a uvial/ lacustrine bog succession may show di erences in organic in ux, which is re ected by the productivities of di erent biota (e.g. land plants vs algae). It is possible that such variations in the gross composition of the organic matter result in little, or no di erence in the molecular compositions and bulk or fraction carbon isotopes (Boreham et al., 1994). However, such di erences in organic matter input may be re ected in the isotopic composition of speci c compounds derived from some speci c organisms thought to be dominant in the depositional environment (Boreham et al., 1994). A greater resolution of this variation in organic matter type will enhance oil-source correlations in coal-bearing basin. Compound-speci c isotope analysis can provide additional information for this purpose at the molecular level (Freeman et al., 1990). Hence, a greater understanding into the source of crude oils may be possible. In this paper, we have utilized the compound speci c stable carbon isotope technique to further enhance the oil-oil and oil-source correlations (e.g. Karlsen et al., 1995; Stoddart et al., 1995). The aim of present study is to determine whether di erent environments of source rock deposition in coal-bearing sequence can be identi- ed from the compound-speci c stable carbon isotopic composition of source extracts and corresponding oils. The approach taken to answer this question is based on the assumption that the organic in ux variations in a uvial/lacustrine bog setting will a ect the stable carbon isotopic compositions of speci c compounds. Here, we concentrate our study on the stable carbon isotopic analysis of individual n-alkanes. The reasons for this selection are: (1) oils derived from coal measures are commonly dominated by para nic hydrocarbons, except for the naphthenic-aromatic oils and condensates found in the Beaufort-Mackenzie basin (Snowdon,1991); and (2) the distributions of n-alkanes from ancient depositional environments have been widely used to infer biological inputs, maturity, and proportions of terrestrial vs marine input (Tissot and Welte, 1984), and the precusor-product relationship has been well established (Peters and Moldowan, 1993). Several studies have utilized the stable carbon isotopic compositions of individual n-alkanes to resolve source of n- alkanes or n-alkane precursors in ancient sediments (Kennicutt and Brooks, 1990; Collister et al., 1994; Boreham et al., 1994). These results indicated that the stable carbon isotopic compositions of individual n- alkanes can be used as a criteria for distinguishing coaland interbedded shale-sourced oils. Oils and related source rocks from the Turpan basin, north-western China have been used to illustrate the application of this methodology. During the past decade, moderate-size accumulations of petroleum have been discovered in the Turpan basin. Most crude oils from the Turpan basin are volatile, light oils with API gravities in the 30±60 range and typically range in color from amber to yellow. n-alkanes are the major class of compounds of the oils, and no evidence can be observed

3 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349± for the oil biodegradation and phase fractionation (Chen and Zhang, 1994). Previous studies showed that the Lower and Middle Jurassic coal-bearing sequences likely act as the primary source for petroleum. Two types of potential source rocks have been identi ed within this section: (1) coal measures from Lower Jurassic Badaowan Formation and Middle Jurassic Xishanyao Formation, mainly deposited under limnic condition; and (2) Middle Jurassic mudstones (Qiketai Formation) deposited in a brackish lacustrine environment. Except the Qiketai lacustrine sediments containing type II organic matter, the organic matter in Badaowan and Xishanyao Formations is attributed to humic material (type III) (Huang et al., 1991; Chen and Zhang, 1994, Hendrix et al., 1995; Sun, 1999). Although there is little doubt that oils discovered in the Turpan basin are derived mainly from the Lower and Middle Jurassic sequence, attempts to characterize a coal- or associated shale-sourced oil have proven unsuccessful. In this paper, we report the preliminary results from the identi cation of coal- and interbedded shale-sourced oils using stable carbon isotopic compositions of n-alkanes from n-c 10 to n-c Analytical procedures Oil and source rock sample information is listed in Tables 1 and 2, respectively, and sample locations are shown in Fig. 1. Sediments were soxhlet extracted with chloroform for approx. 72 h. Asphaltenes were removed from the total chloroform extracts and oils by precipitation with n-hexane followed by ltration. The deasphalted extracts and oils were then separated into saturate, aromatic, and polar (NSO) fractions by column chromatography, using hexane, benzene, and alcohol as the solvents, respectively Gas chromatography Aliphatic hydrocarbon fractions were analyzed by GC using a Hewlett-Packard 5890II Gas Chromatograph equipped with a 25 m0.25 mm i.d. fused silica capillary column coated with a 0.17 mm lm of the HP-1 phase. The temperature program was from 40 C (for oil) or 80 C (for source extracts), held isothermally for 2 min and then increased to 290 C at a rate of 4 C/min, followed by a 30 min hold at 290 C. The carrier gas was nitrogen with a ow rate of 1.0 ml/min Gas chromatography±mass spectrometry GC±MS analyses were conducted on a Voyger 1000 mass spectrometer coupled to a Carlo Erba GC 8000 TOP gas chromatograph with an on-column injection system. Chromatographic separations were performed with a 30 m0.32 mm i.d. fused silica capillary column coated with a 0.25 mm lm of the CB-5 phase (Chrompack). The oven temperature program was from 55 to 205 Cat15 C/min and maintained for 5 min, and then from 205 to 300 Cat2 C/min, followed by a 30 min hold. Helium was used as carrier gas with a ow rate of 1.0 ml/min. The transfer line temperature was 250 C, and the ion source temperature was 200 C. The ion source was operated in the electron impact (EI) mode at an electron energy of 70 ev. Relative amounts of biomarker compounds were determined by multiple ion detection. Table 1 Jurassic oil sample information and bulk composition a Sample no. Oil eld Depth (m) Formation API Wax (%) Sats (%) Arom (%) Res (%) Asph (%) Oil group (%) TH01 Hongtai 1394±1400 Sanjianfang A TH02 Sakeshang ± Xishanyao A TH03 Qiketai Oil mine Sanjianfang n.d. n.d A TH04 Hongfu ±2188 Sanjianfang A TH05 Wenjisang 2828± Sanjianfang n.d. n.d A TH06 Qiuling-shanshan 3236± Xishanyao A TH07 Bake 2379±2391 Xishanyao A TH08 Shanle 2692±2737 Xishanyao A TH09 Hongnan ± Sanjianfang A TH10 Lianmoxin 3372±3379 Xishanyao A TH11 Shengnan 2320± Sanjianfang A TH12 Shengbei 2953± Kalaza B TH13 Shengjinkou? Sanjianfang n.d. n.d B TH14 Pubei ± Qiketai B TH15 Shenqie 2489±2493 Sanjianfang B a API=API gravity, Sats=saturates, Arom=aromatics, Res=resins, Asph=asphaltenes, n.d.=not determined. Saturates, aromatics, resins, and asphaltenes are in normalized percent. Oil group=interpreted genetic oil families (see text for discussion).

4 1352 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349±1362 Table 2 Jurassic source rock description, TOC, rock-eval pyrolysis and vitrinite re ectance data a Sample no. Oil eld Lithology Depth (m) Formation TOC (wt.%) HI (mg HC/org C) S1 (mg/g) S2 (mg/g) Tmax (C) Ro range (%) Ro average (%) Number of vitrinite particles TH16 Hongfu Coal Xishanyao ± TH17 Wenjishang Coal 3877 Xishanyao ± TH18 Sakeshang Coal 3175 Xishanyao ± TH19 Bake Coal Xishanyao ± TH20 Lianmoxin Shale Xishanyao n.d. n.d. n.d. TH21 Hongnan Shale Xishanyao ± TH22 Lianmoxin Shale Qiketai ± TH23 Qiuling-shanshan Shale Qiketai n.d. n.d. n.d. TH24 Shengbei Shale Qiketai 1.07 n.d. n.d. n.d. n.d. n.d. n.d. n.d. a n.d.=not determined. Fig. 1. Map showing location of oil elds from which samples were obtained for geochemical analysis Gas chromatography±isotope ratio mass spectrometry The GC±IRMS analyses were performed on a VG Isochrom II system interfaced to a Hewlett-Packard 5890 gas chromatograph. The GC was tted with a fused silica column (OV1, 50 m0.32 mm i.d.) leading directly into the combustion furnace. Initial temperature of the GC oven was 70 C and held isothermally for 5 min, and then programmed to 290 Cat3 C/min, followed by a 30 min hold. Helium (12 psi) was used as a carrier gas. The injection of samples was conducted in split mode at a ratio of 35:1. The isotope values were calibrated against the reference gas and are reported in the usual ``del'' notation relative to PDB Controls on the instrument accuracy and data quality The accuracy of the instrument was tested daily before sample analysis, by analyzing a mixture of n- alkanes and isoprenoid alkanes with known d 13 C values. The reproducibility, run at di erent times, was typically less than 0.3%. For calibration, a CO 2 reference gas, which is calibrated against the Charcoal Black (a national calibrated standard, with a value of-22.43% based on the PDB standard), was automatically introduced into the IRMS in a series of pulses before and after the array of peaks of interest (Fig. 2). There are two problems in GC±IRMS analysis. One is integrating the isotopic distribution across single-component GC peaks (Hayes et al., 1990); while the other is associated with the identi cation and the correction required for coeluting, unresoved baseline components (Collister et al., 1992). The solution of these problems is achieved using background subtraction software supplied with the Isochrom II instrument. This can be performed automatically or manually. Saturated fractions analyzed in this study are dominated by n-alkanes with a carbon distribution ranging from n-c 10 to n-c 30, and have a low level background matrix or UCM. Fig. 2 shows the typical m/z 44 mass chromatograms of oils and source extracts. As seen in

5 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349± Fig. 2. Typical m/z 44 mass-chromatograms of crude oil (a) and source extracts (b) by GC±IRMS analysis. Fig. 2, n-alkanes are well resolved from neighboring peaks and no major co-elution is observed. For these characteristics, comparison of the measured isotope values for the n-alkane components in the saturated fractions and isolated from the molecular sieve and urea adducted sample fractions revealed a very close match, most components however, agreed to within 0.3±0.5% on average, a value considered well within the instrument precision error (Wilhelms et al., 1994; Ellis and Fincannon, 1998), thus allowing accurate isotopic analyses. Alkanes with more than 29 carbons commonly coelute with compounds in the ``biomarker range'' (e.g. steranes and hopanes) and are therefore, not analyzed isotopically in this paper. 3. Results and discussion 3.1. Turpan basin oil characterization Bulk properties Bulk geochemical data for the oils are given in Table 1. All have a predominance of saturated hydrocarbons relative to aromatics and non-hydrocarbons. API gravities range from 37 to 52 at reservoir depths of 1394±3379 m, which suggests that the reservoired petroleum is dominated by light oil. There is very little correlation between API gravity and depth. An extremely wide variation in the wax content is typical among these crude oils, which ranges from 3.91 to 27.7%. This variation is most likely attributed to variations in coaly source rock organic matter (Isaksen et al., 1998) since phase fractionation (Thompson, 1987) and biodegradation are of no consequence in the Turpan basin (Chen and Zhang, 1994). We did not determine sulfur content of the oils, but most of Turpan basin oils have a low sulfur content (often <0.1 wt.%) according to a previous study (Huang et al., 1991). The low sulfur contents in the oils are consistent with the proposed nonmarine e ective source rocks in the Turpan basin (Huang et al., 1991; Chen and Zhang, 1994; Hendrix et al., 1995; Sun, 1999) Molecular properties Gas chromatography data from the analyses of saturated hydrocarbons of the oils are listed in Table 3. Most Turpan basin oils contain a typical n-alkane distribution with a predominance of a lower carbon number range (n-c 10 ±n-c 21 ), and the ratio of C 21 +/C 21 - ranging from 0.26 to This is in agreement with the API gravity of the oils. Oils produced from various oil elds have variable pristane/phytane (Pr/Ph) ratios (2.67±6.60) indicating terrestrial organic matter input under oxic-suboxic conditions (Peters and Moldowan, 1993; Powell and Boreham, 1994). In all of the oils, the Pr/n-C 17 and Ph/n-C 18 ratios are less than 1.0 except for the oil produced from Sakeshang oil eld, which has a Pr/n-C 17 ratio of The CPI values in the range of n- C 22 ±n-c 30 varies from 1.06 to 1.14, suggesting relatively low oil maturities. There is no correlation between Pr/ Ph ratios and CPI values. Representative mass chromatograms showing the molecular distribution of steranes (m/z 217) and terpanes (m/z 191) for oils are shown in Fig. 3. The proportions of individual classes of sterane compounds (C 27 ±C 29 ) among the Turpan basin oils exhibit

6 1354 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349±1362 Table 3 Geochemical results for crude oils from Turpan basin a Hopanoids Steroids Sample no. Pr/Ph Pr/nC17 Ph/nC18 CPI C21+/C21-22S/22R Ts/Ts+Tm GI C30*/C29Ts 20S/20R bb C27% C28% C29% TH TH TH TH TH TH TH TH TH TH TH TH TH TH TH a Pr/Ph=pristane/phytane; 22S/22R=C31 homohopane 22S/(22S+22R); Ts/(Ts+Tm)=18a(H)-trisnorneohopane/[18a(H)-trisnorneohopane +17a(H)-trisnorhopane]; GI=gammacerane index (gammacerane/hopane100); C30*/C29Ts=17a(H)-diahopane/18a(H)-30-norneohopane; 20S/20R =20S/(20S+20R) C29 14a(H), 17a(H) steranes; bb =C29 20R 14b(H),17b(H)/[14b(H),17b(H)+14a(H),17a(H)]; C29, C28, C27=normalized percentages in sum of C27-C29 steranes; CPI=2 (nc23+nc25+nc27+nc29)/[nc22+2 (nc24+nc26+nc28)+nc30]; C21+/C21-=(sum of the n-alkanes carbon number less than 21)/(sum of the n-alkanes carbon number more than 21). considerable variability, but all with a predominance of C 29 ethylcholestane. Additionally, almost all of the oils have a relative high amount of diasteranes, and the ratio of diasteranes/steranes ranges from 0.45 to The pentacyclic triterpanes in the Turpan basin oils are mainly from the common hopane series. A signi cant feature of the m/z 191 mass chromatograms is the abundance of 17a(H)-diahopane and 18a(H)-30-norneohopane. Other terrigenous triterpanes such as 18a(H)- oleanane and ursane have not been detected, which is consistent with the Jurassic age of the sediments (Grantham et al., 1983; ten Haven and Rullkotter, 1988). Gammacerane is also present, but varies considerably in concentration among samples. There are minor amounts of tricyclic terpanes (C 19 ±C 26 ) in the oils although their intensities in the m/z 191 mass chromatograms are weak and usually clear mass spectra are not available Thermal maturity Several molecular maturation parameters from GC and GC/MS analyses were calculated to estimate relative thermal maturities of the oils (Table 3). We have not attempted to relate these relative maturity levels to speci c levels of maturation that could be measured in their source rocks. Our objective is to assign relative maturites and provide a general framework for interpreting the genetic oil families. Ratios of pristane/n-c 17 and phytane/n-c 18 can be used to evaluate maturity as well as depositional environment. Maturity can be assessed where the oils come from one source rock facies (Connan, 1980). Di erent maturity trends may indicate di erences in depositional environment (Le Tran and Philippe, 1993). Fig. 4 reveals that all of the analyzed oils follow two maturity trends, which indicates that there are probably two main source rock types. The trend A!B is interpreted as lacustrine-sourced oils with relative low pristane/phytane ratios (2.67±3.70), and the trend A!C probably re ects oils sourced from coal measures. The latter have a larger content of land plant organic matter and higher pristane/phytane ratios than the trend A!B (see later discussion). The di erence cannot be due to maturity alone and it will be argued that these oils are from different source facies of the Lower to Middle Jurassic sequence. A better understanding of the oil maturities can be seen using hopane and sterane maturity parameters. The 22S/(22S+22R) ratios for the C 31 17a(H)-homohopanes in the Turpan basin oils are at equilibrium having a range from 0.58 to 0.60 with an average of 0.59 (Table 3), and indicate maturities at or past the onset of oil generation (Seifert and Moldowan, 1986). Typical sterane maturity parameters such as the 20S/(20S+20R) ethylcholestane ratio and the bb/(bb+aa) ethylcholestane ratio are cross-plotted in Fig. 5. C 29 steroid isomerization ratios 20S/(20S+20R) are in the range of 0.31±0.44, less than the pseudo-equilibrium value of 0.55 (Seifert and Moldowan, 1986). The low isomerization ratios for these oils indicate relatively low levels of thermal stress, and below an e ective maturity of 0.9% vitrinite re ection. The trend due to maturity shown in Fig. 5 suggests two oil families: Group A and B. Group

7 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349± Fig. 3. Biomarker fragmentograms for m/z 191 hopane, m/z 217 sterane scans for typical Group A and B oils. A oils plot fairly close together, while Group B oils have lower maturity levels than Group A. These two oil families are considered to correlated with the Xishanyao Formation and Qiketai Formation respectively (see later discussion). Although Ts/(Ts+Tm) ratio has been used successfully as an indicator of thermal maturity for oils and source rocks (Seifert and Moldowan, 1978), it has also been shown to have a strong source in uence (Moldowan et al., 1986). Fig. 6 shows the correlation of the 20S/(20S+20R) ethylcholestane ratios and Ts/ (Ts+Tm) ratios. With increasing maturity, an inverse trend is observed. Therefore, the Ts/(Ts+Tm) ratio of Turpan basin oils is signi cantly in uenced by organic facies rather than by thermal history. Fig. 4. Pristane/n-C 17 ratio vs phytane/n-c 18 ratio calculated from FID gas chromatograms of saturated hydrocarbons of oils Source rock and oil families The thermal maturity di erence between the oils suggests that there is more than one source kitchen. Molecular parameters of source rocks are summarized in

8 1356 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349±1362 Fig. 5. C 29 bb/(bb+aa) vsc 29 aaa 20S/(20S+20R) of crude oils, calculated from MID, GC±MS analysis. *, Group A oils; &, group B oils. Fig. 6. C 29 aa 20S/(20S+20R) vs Ts/(Ts+Tm) ratio from Turpan basin oils, calculated from m/z 191 and 217 fragmentograms in the MID, GC±MS analysis. Table 4. Fig. 7 plots the relative abundance of the C 27 ± C 28 ±C 29 steranes of oils and source extracts, and clearly demonstrates that the oils/source rocks separate into two populations, and support the distinctions made above. Oils from Group A are well correlated with the Xishanyao coal and interbedded shale samples, whereas the Group B oils are well correlated with the Qiketai source rocks. All oils are characterized by a high concentrations of C 29 steranes, indicating a land-plant source (Huang and Meinshein, 1979; Mackenzie et al., 1983). Of signi cance, however, are the variations of C 29 - vs C 27 -steranes of oils and related source rocks (Tables 3 and 4). It is noted that the Group B oils contain lower C 29 sterane and higher C 27 sterane contents than the Group A oils, indicating that depositional environments associated with the source inputs show considerable variablility in this two source kitchens. Previous studies show that the Qiketai Formation is deposited in a semi-deep lacustrine facies (Huang et al., 1991; Wang et al., 1994), and the Xishanyao Formation represents a series of uvial/lacustrine and swamp deposits (Wang et al., 1994; Hendrix et al., 1995). The relative contribution of land plants to these two facies is clearly evident in Fig. 7, with relative C 29 ethylcholestane concentration increasing from the lacustrine facies (Qiketai Formation) to the swamp deposits (Xishanyao Formation). At the same time, the in uence of watercolumn biota (measured from C 27 cholestane) becomes evident in lacustrine facies (Qiketai Formation). Further evidence for the di erentiation of Turpan basin oils comes from the crossplot of the ratio of 18a(H)-trisnorneohopane/(18a(H)-trisnorneohopane+17a(H)-trisnorhopane) [Ts/(Ts+Tm)] against the 17a(H)-diahopane/18a(H)-30-norneohopane [C 30 */ C 29 Ts] (Fig. 8). Although there is possible in uence of thermal maturity on the Ts/(Ts+Tm) and C 30 */C 29 Ts ratios, the depositional environment appears to exert the principal control on the relative Ts and Tm concentrations in this study as discussed above, and it can be used as a source-dependant parameter. On the other hand, the relative amounts of 17a(H)-diahopane and 18a(H)-30-norneohopane also depend most strongly on depositional environments, and the C 30 */C 29 Ts ratios can be considered to be a function of Eh/pH in the depositional environments. High C 30 */C 29 Ts ratios indicate that source rocks were deposited under oxicsuboxic conditions (Moldowan et al., 1991; Peters and Moldowan, 1993). In contrast with the Group A oils and related Xishanyao source rocks, Group B oils and related Qiketai source extracts have higher Ts/(Ts+Tm) ratios and lower C 30 */C 29 Ts ratios. It suggests that the Xishanyao source rocks accumulated under more oxic conditions compared to the Qiketai source rocks. This is also supported by variations in pristane/phytane and gammacerane index (Fig. 9). The depositional environments for Xishanyao and Qiketai Formations, inferred from biomarker assemblages of oils, are in accordance with the result of sedimentary facies analysis (Wang et al., 1994). Of signi cance in Fig. 8, however, is the more scattered distribution of Group A oils compared to the Group B oils due to the wider range of C 30 */C 29 Ts ratios (0.70±1.77) (Table 3). Although application of the C 30 */C 29 Ts parameter has not been well established, the similar maturities of Group A oils (Fig. 5) suggests that the e ects of thermal maturity on C 30 */C 29 Ts ratios is small. An alternative explanation can be attributed to the heterogeneous nature of depositional environment in coal measures which results in the changeable redox potential of the source sediments, and thereby having a signi cant e ect on the C 30 */C 29 Ts values. The tighter cluster of Group B oils indicates the relative homogeneous nature of source rock depositional environment. Again, these ndings are in agreement with the

9 Table 4 Biomarker data for source rocks from Turpan basin a Y. Sun et al. / Organic Geochemistry 31 (2000) 1349± Hopanoids Steroids Sample no. Pr/Ph Pr/nC17 Ph/nC18 22S/22R Ts/Ts+Tm GI C30*/C29Ts C29Ts/C29ab 20S/20R bb C27% C28% C29% TH TH TH TH TH TH TH TH TH a Pr/Ph=pristane/phytane; 22S/22R=C31 homohopane 22S/(22S+22R); Ts/(Ts+Tm)=18a(H)-trisnorneohopane/[18a(H)-trisnorneohopane +17a(H)-trisnorhopane]; GI=gammacerane index (gammacerane/hopane100); C30*/C29Ts=17a(H)-diahopane/18a(H)-30-norneohopane; C29Ts/C29ab=18a(H)-30-norneohopane/17a(H)-30-norhopane; 20S/20R=20S/(20S+20R) C29 14a(H), 17a(H) steranes; bb=c29 20R 14b(H),17b(H)/[14b(H),17b(H)+14a(H),17a(H)]; C29, C28, C27=normalized percentages in sum of C27±C29 steranes. Fig. 7. Ternary diagram showing the relative abundances of C 27, C 28, and C 29 regular steranes in the saturated hydrocarbons of oils and source extracts determinded by MID, GC± MS analysis. *, Group A oils; &, group B oils; ^, source rocks from Xishanyao Fm; ~, source rocks from Qiketai Fm. Fig. 8. C 30 */C 29 TsvsTs/(Ts+Tm) ratio calculated from m/z 191 fragmentograms in MID, GC±MS analysis. *, Oils from group A; &, oils from group B; *, source rocks from Xishanyao Fm; &, source rocks fom Qiketai Fm. descriptions of source rock depositional environments (Wang et al., 1994) Compound speci c isotope analysis It has been shown that the traditional molecular organic geochemical parameters have separated the Turpan basin oils into two families. However, similar maturities and a broad range for some source-dependant parameters suggests heterogenous source facies for Group A oils. Based on the oil-oil and oil-source correlations, Group A oils are considered to be derived from coal measures. Therefore, both coals and associated shales are possible sources for the oils. In order to Fig. 9. Variations in Pr/Ph and gammacerane index for the Turpan basin oils. GI=gammacerane index (gammacerane/ hopane100). *, Group A oils; &, group B oils.

10 1358 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349±1362 Table 5 d 13 C values (%) for n-alkanes of the oils and source extracts from Turpan basin a ID TH01 TH02 TH03 TH04 TH05 TH06 TH07 TH08 TH09 TH10 TH11 TH16 TH17 TH18 TH19 TH20 TH21 n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c n-c a n-cx=n-alkane of carbon number X. attempt a di erentiation of coal-sourced oils and interbedded shale-sourced oils, we utilized the GC/IRMS technique to gain a better understanding of the source variations of the Group A oils. Stable carbon isotopic compositions for n-alkanes from oils and source extracts are summarized in Table 5. n-alkane isotope pro les for the source extracts are plotted in Fig. 10, and two distinct isotope curves can be identi ed. Shales show a trend toward isotopically lighter values with increasing n-alkane chain length (negative slope) with an isotopic range being typically 2± 3%. This phenomenon has been observed previously and is a characteristic of terriestrial organic matter and related oils (Bjorùy et al., 1991; Dzou and Hughes, 1993; Wilhelms et al., 1994; Murray et al., 1994). Coals show that n-alkane isotope curves are relatively at across the range. This trend is a characteristic of a homogenous precursor for the derivation of n-alkanes. Under such circumstances there is typically not much of an isotopic di erences between long and short chain alkanes (Eglinton, 1994; Murray et al., 1994). Alternatively, n-alkanes, particularly those with carbon number less than 21, could be derived from algae that happen to produce isotopically similar alkanes as the land plants (Boreham et al., 1994). In addition to chain length dependant isotope e ects, there are dramatic changes in the n-alkane isotope ratios in the range n- C 20 ±n-c 29 for coal and interbedded shale extracts. n- Alkanes from coal extracts are enriched in 13 C relative to those of interbedded shale extracts. Carbon isotope excursions are in the range of 2±3%. This nding is in Fig. 10. Range of isotope values of n-alkanes in coal and interbedded shale samples. *, Coals; ^, inerbedded shales. contrast to that of a study of the Permian Toolachee coal measures from Cooper and Eromanga basin, Australia, where the n-alkanes isotopic pro les of shale extracts is heavier than that of coal extracts (Boreham and Summons, 1999), and highlights the value of sitespeci c geochemical analysis. In view of the similar maturities of source rocks (Table 2), cracking and other secondary processes cannot have signi cantly in uenced the n-alkane isotope ratios (Bjorùy et al., 1992; Clayton and Bjorùy, 1994 ). Since vitrinite re ectance values for most studied samples suggest that these source rocks are close to or at the onset of the conventional oil window (Ro% 0.5±0.7), there may be a di erence between the proportion of free

11 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349± n-alkanes (bitumen) and those released from the kerogen, which will be generated by the di erent chemical kinetics between the coal and shale (Collister et al., 1992; Boreham et al., 1994; Murray et al., 1994). However, the di erence of isotope pro les for coal and interbedded shale extracts has to be mostly due to the source input variations. The source of organic matter in the sedimentary system include predominantly algae and the remains of terrestrial vegetation. Terrestrial organic matter exhibits varying d 13 C values, depending on the photosynthetic pathways used by the plants. The observed range of d 13 C values for C3 plants is about- 23% to-34% with an average value of about-27% (Smith and Epstein, 1971; Benedict, 1978). Our result is in generally agreement with C3 photosynthetic pathways. Planktonic algae can have d 13 C values more negative than those of terrestrial plants due to the high concentrations of dissovled CO 2 in uvial/lacustrine bog systems (Herczeg and Fairbanks, 1987; Hollander and Mckenzie, 1991; Mayer and Schwark, 1999). Thus, more negative ratios are predicted if algal productivity prevails in the sedimentary record, whereas more positive d 13 C values can be indicative of input from predominantly terrestrial biomass. A study of the Green River oil shale revealed that the isotopic compositions of the n-alkanes are di erent between algae-sourced organic matter and terrestrial plant-sourced organic matter. Comparatively, n-alkanes derived from algae are more depleted in 13 C (Collister, 1992). Considerable organic in ux variations, particularly algae vs land plants, are likely to take place in a uvial/lacustrine bog setting. We attributed a high proportion of land plant organic matter for the coals, and greater contents of algal material for the interbedded shales. Hence, source-dependant isotope e ects are expected to cause an enrichment of 13 Cforn-alkanes extracted from coals with respect to that of interbedded shales. The data in Fig. 10 suggests that the shapes of the n- alkane isotope pro les from coals and interbedded shales have the potential to di erentiate the coalsourced oils from interbedded shale-sourced oils. Stable carbon isotopic compositions of the n-alkanes from n- C 10 to n-c 29 of Group A oils are determined (Table 5). Because there is no signi cant di erence between Fig. 11. n-alkane isotope pro les of coal-sourced oils in Turpan basin. *, Coals; +, interbedded shales; ----, oil TH01; Ð, oil TH02; ± - ±, oil TH07. Fig. 13. n-alkane isotope pro les of mix-sourced oils from Wenjishang and Honghu oil elds. *, Coals; +, interbedded shales; Ð, oil TH04, ----, oil TH05. Fig. 12. n-alkane isotope pro les of interbedded shale-sourced oils in Turpan basin. *, Coals; +, interbedded shales; Ð, oil TH10; ± - ±, oil TH06; ----, oil TH08. Fig. 14. n-alkane isotope pro les of mix-sourced oils from Hongnan and Shengnan oil elds. *, Coals; +, interbedded shales; ----, oil TH09; Ð, oil TH11.

12 1360 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349±1362. Group B oils, derived from a lacustrine source rock of a suboxic, brackish environment with organic matter bearing a relatively high amount of plankton and low amounts of land-plant input (Qiketai Formation). Fig. 15. n-alkane isotope pro les of mix-sourced oils from Qiketai oil mine. *, Coals; +, interbedded shales; Ð, oil TH03. maturity of oils and source rocks based on the maturity parameters derived from biomarkers (Tables 3 and 4), oil-source correlations can be undertaken using n-alkane isotopic curves of oils and sediments, and shown in Figs. 11±15. Most oils are characterized by a n-c 10 ±n-c 29 n- alkane isotopic compositions with a range from-28 to- 24%, yet show wide variations in n-alkane isotope pro- les. In view of the fairly similar maturities of Group A oils (based on the various maturity parameters discussed above, see Table 3 and Fig. 5), this change is considered to be due to the variation of source input/depositional setting. The oils from Sakeshang oil eld, Bake oil eld and Hongtai oil eld, have similar isotopic pro les to the coal extracts (Fig. 11), indicating these oils are mainly generated from coal seams. In contrast, the oils from Qiuling-shanshan oil eld, Lianmoxin oil eld and Shanle oil eld, show isotopic patterns were closely related to the shale extracts (Fig. 12), indicating these oils are mainly derived from shales. The remaining Group A oils, however, are believed to be of mixed source from both coals and interbedded shales (Figs. 13± 15). The n-c 20 ±n-c 29 n-alkanes of oils from Wenjishang oil eld and Hongfu oil eld, are slightly more depleted in 13 C than that of coal extracts (Fig. 13), which could be interpreted as a considerable contribution from interbedded shales. Other mix-sourced oils include the Hongnan oil eld, Shengnan oil eld (Fig. 14) and Qiketai oil mine (Fig. 15). 4. Summary Selected oil samples from Turpan basin have been characterized by organic geochemistry techniques and correlated to their source rocks. Two major oil families are di erentiated:. Group A oils, derived from coal measures deposited in a strong oxic environment with organic matter from higher land plants (Xishanyao Formation). Compound-speci c isotope analysis has been used to further di erentiate coal-sourced and interbedded shalesourced Group A oils. Based on the carbon isotopic pattern of n-alkanes from source rock extracts and oils in the Turpan basin, some important conclusions are drawn: (i) coal-sourced oils are mainly present in Bake, Sakeshang and Hongtai oil elds, (ii) shale-sourced oils occur in the Qiuling-shanshan, Lianmoxin and Shanle oil elds and, (iii) mixed oils were found in Wenjisang, Honghu, Shengnan, Hongnan oil elds and Qiketai oil mine. Acknowledgements We are grateful for helpful advice from Professors Dehan Liu and Jiyang Shi. Highly constructive reviews were received from Drs. C. J. Boreham, G. H. Isaksen and S. Inan, to whom we are most grateful. P.P. thank the funding from the Chinese Academy of Sciences for ``The Hundred Talents''. Financial support for this work was provided mainly by Chinese Academy of Sciences (Grant No. KZ951-B1-412). References Benedict, C.R., The fractionation of stable carbon isotopes in photosynthesis. What's New in Plant Physiol 9, 13± 16. Bjorùy, M., Hall, K., Gillyon, P., Jumeau, J., Carbon isotope variations in n-alkanes and isoprenoids in whole oils. Chem. Geol. 93, 13±20. Bjorùy, M., Hall, P.B., Hustad, E., Williams, J.A., Variation in stable carbon isotope ratios of individual hydrocarbons as a function of arti cial maturity. Org. Geochem. 19, 89±105. Boreham, C.J., Summons, R.E., New insights into the active petroleum systems in the Cooper and Eromanga basin, Australia. APPEA Journal 39 (1), 263±296. Boreham, C.J., Summons, R.E., Roksandic, Z., Dowling, L.M., Chemical, molecular and isotopic di erentiation of organic facies in the Tertiary lacustrine Duaringa oil shale deposit, Queensland, Australia. Org. Geochem. 21, 685±712. Brooks, J.D., The use of coals as indicators of the occurrence of oil and gas. APEA Journal 10, 35±40. Chen, K., Zhang, Z., The studies on coal-derived oils in Turpan-Hami basin, China (in Chinese). Chinese Science (B) 24, 1216±1222. Clayton, C.J., Carbon isotope fractionation during natural gas generation from kerogen. Marine and Petroleum Geology 8, 232±240.

13 Y. Sun et al. / Organic Geochemistry 31 (2000) 1349± Clayton, C.J., Bjoroy, M., E ect of maturity on 13 C/ 12 C ratios of individual compounds in North Sea oils. Org. Geochem. 21, 737±750. Collister J.W., An isotopic biogeochemical study of the Green River oil shale (Piceance Creek Basin, Colorado). Ph.D thesis, Indiana University. Collister, J.W., Lichtfouse, E., Hiesshima, G., Hayes, J.M., Partial resolution of sources of n-alkanes in the saline portion of the Parachute Creek Member, Green River Formation (Piceance Creek Basin, Colorado). Org. Geochem. 21, 645±659. Collister, J.W., Summons, R.E., Lichtfouse, E., Hayes, J.M., An isotopic biogeochemical study of the Green River oil shale. Org. Geochem. 19, 265±276. Connan J., Biological markers in crude oils. In: Mason, J.F. (Ed.), Petroleum Geology in China. Pennwell, London, Chapter 4, pp Curiale, J.A., Lin, R., Tertiary deltaic and lacustrine organic facies: comparison of biomarker and kerogen distributions. Org. Geochem. 17, 785±804. Curry D.J., Emmett J.K., Hunt J.W., Geochemistry of aliphatic-rich coals in the Cooper Basin, Australia and Taranaki Basin, New Zealand: implication for the occurrence of potentially oil-generative coals. In: Scott A.C., Fleet A.J. (Eds.), Coal and Coal-bearing Strata as Oil-prone Source Rocks? Geological Society, London, Special Publication Vol. 77, pp Dzou, L.I.P., Hughes, W.B., geochemistry of oils and condensates, K eld, o shore Taiwan: a case study in migration fractionation. Org. Geochem. 20, 437±462. Eglinton, T.I., Carbon isotopic evidence for the origin of macromolecular aliphatic structures in kerogen. Org. Geochem. 21, 721±736. Ellis, L., Fincannon, A.L., Analytical improvements in irm-gc/ms analysis: Advanced techniques in tube furnace design and sample preparation. Org. Geochem. 29, 1101± Freeman, K.H., Hayes, J.M., Trendel, J.M., Albrecht, P., Evidence from carbon isotope measurements for diverse origins of sedimentary hydrocarbons. Nature 343, 254±256. Gordon, T.L., Talang Akar coals Ð Ardjuna subbasin oil source. Proc. 14th IPA Ann. Conv. 2, 91±120. Grantham, P.J., Posthuma, J., Baak, A., Triterpanes in a number of far-eastern crude oils. In: Bjoroy, M. et al. (Eds.), Advance in Organic Geochemistry J. Wiley and Sons, New York, pp. 675±683. Hayes, J.M., Freeman, K.H., Popp, B.N., Hoham, C.H., Compound speci c isotope analysis: a novel tool for reconstruction of ancient biochemical process. Org. Geochem. 16, 1115±1128. Hendrix, M.S., Brassel, S.C., Carroll, A.R., Graham, S.A., Sedimentology, organic geochemistry, and petroleum potential of Jurassic coal measures: Tarim, Junggar, and Turpan basins, Northwest China. AAPG Bulletin 79, 929±959. Herczeg, A.L., Fairbanks, R.G., Anomalous carbon isotope fractionation between atmospheric CO 2 and dissolved inorganic carbon induced by intense photosynthesis. Geochim. Cosmochim. Acta 52, 895±899. Hollander, D.J., Mckenzie, J.A., CO 2 control on carbonisotope fractionation during aqueous photosynthesis: a paleo-pco 2 barometer. Geology 19, 929±932. Huang, D., Zhang, D., Li, J., Huang, X., Hydrocarbon genesis of Jurassic coal measures in the Turpan basin, China. Org. Geochem. 17, 827±838. Huang, W.Y., Meinschein, W.G., Sterols as ecological indicators. Geochim. Cosmochim. Acta 43, 739±745. Hunt, J.M., Generation of gas and oil from coal and other terrestrial organic matter. Org. Geochem. 17, 673±680. Inan, S., Namik Yalcin, M., Mann, U., Expulsion of oil from petroleum source rocks: inferences from pyrolysis of samples of unconventional grain size. Org. Geochem. 29, 45± 61. Isaksen G.H., Organic geochemistry of paleodepositional environments with a predominance of terrigenous higherplant organic matter. In: Huc A.Y. (Ed.), Paleogeography, Paleoclimate, and Source Rocks. AAPG Studies in Geology No. 40, pp. 81±104. Isaksen, G.H., Curry, D.J., Yeakel, J.D., Jenssen, A.I., Controls on the oil and gas potential of humic coals. Org. Geochem. 29, 23±44. Karlsen D.A., Nyland B., Flood B., Ohm S.E., Brekke T., Olsen S., Backer-Owe K., Petroleum Geochemistry of the Haltenbanken, Norwegian Continantal Shelf. In: Cubitt, J.M., England, W.A. (Eds.), The Geochemistry of Reservoirs Geological Society, London, Special Publication 86, pp. 203± 254. Katz, B.J., Kelley, P.A., Royle, R.A., Jorjorian, T., Hydrocarbon products of coals as revealed by pyrolysis-gas chromatography. Org. Geochem. 17, 711±722. Kennicutt, M.C., Brooks, J.M., Unusual normal alkane distributions in o shore New Zealand sediments. Org. Geochem. 15, 193±197. Killops, S.D., Funndl, R.A., Suggate, R.D., Sykes, R., Peters, K.E., Walters, C. et al., Predicting generation and expulsion of para nic oil from vitrinite-rich coals. Org. Geochem. 29, 1±21. Le Tran, K., Philippe, B., Oil and rock extract analysis. In: Bordenave, M.L. (Ed.), Applied Petroleum Geochemistry. Edition Technip, Paris, pp. 72±394. Mackenzie, A.S., Li, R., Maxwell, J., Moldowan, J.M., Seifert, W.K., Molecular measurements of thermal maturation of Cretaceous shales from the Overthrust Belt, Wyomoing, USA. In: Bjorùy, M. et al. (Eds.), Advance in Organic Geochemistry J. Wiley and Sons, New York, pp. 496±503. Mayer, B., Schwark, L., A 15,000-year stable isotope record from sediments of Lake Steisslingen, Southwest Germany. Chem. Geol. 161, 315±337. Moldowan, J.M., Fago, F.J., Carlson, R.M.K., Young, D.C., Duyne, G.V., Clardy, J. et al., Rearranged hopanes in sediments and petroleum. Geochim. Cosmochim. Acta 55, 3333±3353. Moldowan, J.M., Sundararaman, P., Schoell, M., Sensitivity of biomarker properties to depositional environment and/or source input in the Lower Toarician of S.W. Germany. Org. Geochem. 10, 915±926. Murray, A.P., Summons, R.E., Boreham, C.J., Dowling, L.M., Biomarker and n-alkane isotope pro les for Tertiary oils: relationship to source rock depositional setting. Org. Geochem. 22, 521±542. Peters, K., Moldowan, J.M., The Biomarker Guide: Interpreting Molecular Fossils in Petroleum and Ancient Sediments, Prentice Hall, New Jersey.